On 22/giu/09, at 13:00, Rostislav Khlebnikov wrote:


is there any reason for not making FEFieldFunction::set_active_cell virtual?
I'll explain my use-case and why I need that polymorphic behaviour.
I have an electric potential distribution in volume and I need to calculate Joule heat. It depends on gradient of electric potential, which is a result of finite element calculation. Thus I use FEFieldFunction to calculate those gradients. Plus to that Joule heat also depends on electric conductivity (which is also a FEFieldFunction on the same DoFHandler). And so, I need to propagate set_active_cell to both of these functions (and it would be nice if I could maintain at least FEFieldFunction abstraction). It doesn't seem to be possible without virtual set_active_cell().

There is no particular reason, except the fact that I never needed another level of abstraction.

Still it is not clear to me why you would need that to be virtual. What you need only requires subsequent calls to set_active_cell, as in

joule.set_active_cell(the_cell);
conductivity.set_active_cell(the_cell);

Maybe I don't understand your question...

Btw, FEFieldFunction was thought as a way to interpolate on arbitrary points inside cells. If you are doing this with quadrature points, you are probably better off using FEValues directly...

Luca.

--
Luca Heltai <[email protected]>
http://people.sissa.it/~heltai/
University of Pavia, Dep. of Mathematics
Phone : +39 0382 98 5680, Office: A10
--
There are no answers, only cross references.




_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii

Reply via email to